In the prior art, landmarks that were triangulated using a navigation system and a camera used computationally expensive algorithms such as particle filters or methods such as Simultaneous Localization and Mapping (SLAM) to estimate the landmark position error. Some prior methods simply guessed the position error of triangulated landmarks.
Navigation solutions were commonly estimated using a Global Positioning System (GPS) receiver and an inertial measurement unit (IMU). This has the disadvantage that if GPS is unavailable (as it is in many areas such as in buildings, shopping malls, underground parking structures), then the navigation solution will decay unless the IMU is of a high quality and thus very expensive. Navigation solutions were also estimated using inertial measurement units and digital cameras. The measurements were integrated at the tight level, which means that pixel measurements were combined with accelerometer and gyroscopes measurements from the IMU. In this method, if the system has a poor current estimate of its navigation state, it may enter a failure mode that produces incorrect navigation solutions from which it cannot recover. This can occur as a consequence of having an inexpensive IMU.